| Allgemeine Angaben |
|
| Seminar - Bio-inspired Data Processing and Learning With Neural Networks (IN2107, IN4996) | | |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
 Allocations: 1 | |
| eLearning[Provide new moodle course in current semester] |
|
|
| |
| Angaben zur Abhaltung |
|
There will be a pre-course meeting on the 8th of February at 4 pm, held in zoom. Everybody is free to attend. Attendance is highly recommended.
Topic: Pre-Course Meeting: Seminar - Bio-inspired Data Processing and Learning With Neural Networks
Time: Feb 8, 2021 04:00 PM Amsterdam, Berlin, Rome, Stockholm, Vienna
Join Zoom Meeting
https://tum-conf.zoom.us/j/64171357654?pwd=R0R6Skl5d29LYkhSc0JURGJnYzNmQT09
Machine learning and especially deep learning are valuable tools for a variety of
applications. Sensor data processing is one area where these methods provide
solutions to problems such as object detection, tracking or classification.
Recently, Spiking Neural Networks (SNNs) have emerged as a new generation of neural
networks. They are able to bridge the gap between technology and neuroscience. SNNs
are strongly inspired by findings from neurosciences and offer great potential in terms of
energy efficiency and efficient learning.
Autonomous driving and drone technology are just two of the many areas that could
benefit greatly from these advances.
The main goal of this seminar is the research on the most recent trends in both SNNs
and deep learning, as well as the study of their differences. Moreover, the seminar will
deal with the application of this technology to the processing of sensor data. Each
student will pick a topic from a list, and will need to autonomously write a scientific
report covering its main aspects. Topics will range from specific aspects of deep learning
and SNNs like data encoding and learning algorithms, to their application for processing
sensor data.
In order to be accepted for this seminar, applicants will need to send a brief letter (max. one
page) to the organizers explaining their background and main motivation for attending the
seminar.
Please send this letter in english to the following e-mail address by 17.02.2021:
lopez.randulfe@tum.de
subject: "N2107 - Application"
|
|
|
As the main topic of this seminar is sensor data processing and learning, each student is
advised to have a basic understanding of neural networks and/or sensor data processing.
This can be any sensor (LIDAR, RADAR, vision, ultrasonic, …) as well as any kind of
neural network (MLP, CNN, RNN, SNN, …). Basic knowledge of neuroscience may be
helpful in certain topics.
|
|
|
The goal of this seminar is for the students to learn essential scientific reading, writing and
presentation skills. As part of this seminar each student will be conducting a literature review in
a predefined field. A paper with a maximum length of 6 pages (including bibliography) has to be
handed in at the end.
Generally the student will be asked to perform either a literature review for a specific topic or
do a comparison between different approaches for solving the same problem. At the end of the
semester each student has to hold an oral presentation of 20 minutes including a discussion,
where the work done will be presented.
|
|
|
|
|
|
|
|
|
| Für die Anmeldung zur Teilnahme müssen Sie sich in TUMonline als Studierende*r identifizieren. |
Note: In order to be accepted for this seminar, applicants will need to send a brief letter (max. one
page) to the organizers explaining their background and main motivation for attending the
seminar.
Please send this letter in english to the following e-mail address by 17.02.2021:
lopez.randulfe@tum.de
subject: "N2107 - Application"
|
|
| |
| Zusatzinformationen |
|
|
|
|
|
|
